Download A noninvasive algorithm to exclude pre-capillary pulmonary hypertension

Eur Respir J 2011; 37: 1096–1103
DOI: 10.1183/09031936.00089610
CopyrightßERS 2011
A noninvasive algorithm to exclude
pre-capillary pulmonary hypertension
D. Bonderman*, P. Wexberg*, A.M. Martischnig*, H. Heinzl#, M-B. Lang*,
R. Sadushi*, N. Skoro-Sajer* and I.M. Lang*
ABSTRACT: Current guidelines recommend right heart catheterisation (RHC) in symptomatic
patients at risk of pre-capillary pulmonary hypertension (PH) with echocardiographic systolic
pulmonary artery pressures o36 mmHg. Growing awareness for PH, a high prevalence of postcapillary PH and the inability to distinguish between pre- and post-capillary PH by echocardiography
have led to unnecessary RHCs. The aim of our study was to assess whether standard noninvasive
diagnostic procedures are able to safely exclude pre-capillary PH.
Data from 251 patients referred for suspicion of pre-capillary PH were used to develop a
noninvasive diagnostic decision tree. A prospectively collected data set of 121 consecutive
patients was utilised for temporal validation.
According to the decision tree, patients were stratified by the presence or absence of an
electrocardiographic right ventricular strain pattern (RVS) and serum N-terminal brain natriuretic
peptide (NT-proBNP) levels below and above 80 pg?mL-1. In the absence of RVS and elevated NTproBNP, none of the patients in the prospective validation cohort were diagnosed with pre-capillary
PH by RHC. Combining echocardiography with the diagnostic algorithm increased specificity to
19.3% (p50.0009), while sensitivity remained at 100%.
Employing ECG and NT-proBNP on top of echocardiography helps recognise one false positive
case per five patients referred with dyspnoea and echocardiographic suspicion of PH, while not
missing true pre-capillary PH.
AFFILIATIONS
*Dept of Cardiology, and
#
Core Unit for Medical Statistics and
Informatics, Medical University of
Vienna, Vienna, Austria.
CORRESPONDENCE
D. Bonderman
Dept of Internal Medicine II, Division
of Cardiology
Medical University of Vienna
Währinger Gürtel 18-20
1090 Vienna
Austria
E-mail: diana.bonderman@
meduniwien.ac.at
Received:
June 10 2010
Accepted after revision:
July 28 2010
First published online:
Aug 06 2010
KEYWORDS: Diagnostic procedures, pulmonary hypertension
re-capillary pulmonary hypertension (PH)
is a severe condition leading to right heart
failure and death within 2–3 yrs after
diagnosis, if left untreated [1]. While idiopathic
pulmonary arterial hypertension (PAH) is rare,
associated forms of PAH [2] are more common
and may be triggered by collagen vascular
disease, appetite suppressants [3], HIV-infection
[4], increased shear stress and hypoxia [2, 5–7].
A series of medical conditions, including infection, immune disorders, inflammatory bowel
disease and permanent venous catheters [8, 9],
predispose to chronic thromboembolic pulmonary hypertension (CTEPH) [10].
P
According to international guidelines [2], which
were recently updated [11–13], invasive haemodynamic measurement by right heart catheterisation (RHC) is recommended in patients with
clinical suspicion of pre-capillary PH if systolic
pulmonary artery pressure (Ppa,sys) by transthoracic echocardiography (TTE) is o36 mmHg.
For editorial comments see page 994.
1096
VOLUME 37 NUMBER 5
An invasively measured pulmonary capillary
wedge pressure of 15 mmHg has been used to
discriminate between pre- and post-capillary pulmonary pressure elevation which mostly occurs
as a consequence of left-sided heart disease [14].
The growing awareness for PH, a high prevalence of post-capillary PH [15] and the inability of TTE to distinguish between pre- and
post-capillary PH have necessitated invasive haemodynamic measurements for exclusion. Furthermore, Doppler echocardiography is frequently
inaccurate in estimating Ppa,sys [16]. Given a low
specificity of PH symptoms and Ppa,sys elevations
by TTE, the decision to either proceed with RHC
or withhold further invasive testing is a common
clinical problem, especially in high-volume PH
referral centres [17]. Although rare [18], complications of RHC may be ventricular tachyarrhythmia,
vascular or ventricular perforation, bleeding,
pneumothorax and even death [19]. However,
a restrictive use of RHC may delay a timely
diagnosis and treatment [20].
European Respiratory Journal
Print ISSN 0903-1936
Online ISSN 1399-3003
EUROPEAN RESPIRATORY JOURNAL
D. BONDERMAN ET AL.
Apart from TTE, 12-lead electrocardiography (ECG), serum
N-terminal brain natriuretic peptide (NT-proBNP) and lung
function tests with blood gas analysis have been recommended
in patients who present with dyspnoea. Typical ECG signs of
pre-capillary PH are right ventricular hypertrophy and strain
(RVH and RVS, respectively), and signs of right atrial dilation
such as P-pulmonale. The role of ECG for the diagnosis of PH
has been investigated in a large US registry initiated in the
early 1980s. RICH et al. [21] reported that electrocardiographic
RVH was present in 87% and right axis deviation in 79% of
patients with idiopathic PH. Elevated serum NT-proBNP [22]
and hypocapnia [23] have been established as independent
markers of mortality in PH. However, the combined diagnostic
use of these noninvasive tests has not been validated. Therefore, we tested the ability of standard noninvasive diagnostic
procedures to correctly identify or exclude pre-capillary PH in
patients referred for clinical suspicion of PH and echocardiographic Ppa,sys o36 mmHg. We used a retrospective data
set of 251 consecutive individuals and constructed a simple
noninvasive diagnostic algorithm. In a prospective temporal
validation study enrolling 121 new consecutive patients, the
incremental diagnostic value of the combined use of TTE, ECG
and NT-proBNP over TTE alone was confirmed.
METHODS
Setting and study design
The study was approved by the Ethics Committee of the
Medical University of Vienna (Vienna, Austria), and written
informed consent was collected from prospectively enrolled
patients. The study was performed at the Pulmonary Hypertension Unit of the Medical University of Vienna as a tertiary
referral centre for PH.
Retrospective study
Between January 2002 and April 2007, 462 patients were referred
for the evaluation of PH. TTE, 12-lead ECG, serum NT-proBNP
and lung function tests including arterial carbon dioxide tension
(Pa,CO2) were performed upon admission in each patient.
Patients with echocardiographic Ppa,sys ,36 mmHg (n580) were
not considered for invasive haemodynamic assessment. Patients
carrying pacemakers or implantable cardioverters/defibrillators
(n512) were excluded because of the inapplicability of standard
ECG criteria. 71 patients were excluded because of severe
valvular disease, congenital malformations and/or severely
impaired left ventricular function. Of the 299 patients who were
considered appropriate study candidates, 48 were excluded
because of incomplete data, leaving 251 patients for analysis.
Based on the invasive haemodynamic evaluation by RHC,
the diagnosis ‘‘pre-capillary PH’’ or ‘‘no pre-capillary PH’’
was made.
PULMONARY VASCULAR DISEASE
the study. In the remaining 121 patients, noninvasive diagnostic
procedures were performed in an outpatient setting. Based on
the diagnostic algorithm, each patient was categorised as ‘‘precapillary PH excluded’’ or ‘‘pre-capillary PH likely’’. In a next
step, all patients underwent invasive haemodynamic assessment by RHC. Based on these results, a diagnosis was made that
served as the validation standard.
Transthoracic echocardiography
All TTE studies were performed by board certified physicians in
the echo laboratory of the Medical University of Vienna using
high-end scanners, such as Acuson Sequoia (Siemens AG,
Erlangen, Germany) and Vivid 5 and Vivid 7 (General Electric
Medical System, Milwaukee, WI, USA). The current standard
in our echo lab is that all readings are performed by two
independent observers. TTE studies were based on measurements that are broadly available and routinely used in the
evaluation of patients with suspected PH.
Right ventricular dysfunction was diagnosed based on a visual
assessment and on a tricuspid annular plane systolic excursion
of ,18 mm. Ppa,sys was calculated by adding estimated right
atrial pressure to the tricuspid regurgitation pressure gradient.
No contrast agents were used for enhancement of Doppler
signals. Right atrial pressure was estimated based on the
diameter and respiratory variation of the inferior vena cava [11].
Left ventricular hypertrophy was diagnosed if end-diastolic
septal thickness in the apical four-chamber view was o12 mm.
Left ventricular diastolic dysfunction was diagnosed in the
presence of a restrictive or pseudo-normal filling pattern and
normal or only mildly abnormal left ventricular ejection
fraction (.50%). A restrictive filling pattern was defined by
an E/A waves ratio .2 and a deceleration time ,150 ms.
A pseudo-normal filling pattern was diagnosed if E/A waves
ratio and deceleration time were normal but changed to
abnormal after Valsalva [24].
Prospective study
For temporal validation of the diagnostic algorithm constructed
from the retrospective data set, we enrolled consecutive patients
referred between June 2007 and October 2008. From a total of
222 individuals, 101 were excluded because of valvular heart
disease (n529), left ventricular dysfunction (n58), congenital
heart disease (n58), pacemaker (n517) and an echocardiographic Ppa,sys ,36 mmHg (n530). In one patient NT-proBNP
had not been determined and one patient died prior to a
complete assessment. Seven patients refused to participate in
ECG
Retrospective and prospective study
A 12-lead ECG was recorded according to clinical standards at
a paper speed of 25 mm?s-1. ECGs were analysed by two
cardiologists who were blinded to the clinical and echocardiographic data. ECG rulers and callipers were used. In case of
disagreement, consensus was achieved between the two
observers in a second reading. The following parameters were
obtained: 1) presence or absence of sinus rhythm; 2) heart rate
in beats per min; 3) P-wave amplitude in mV in lead II; 4)
P-wave axis in degrees; 5) presence or absence of P-pulmonale
defined as a P-wave amplitude .0.25 mV; 6) electric heart axis
in degrees; 7) right axis deviation defined as a QRS axis .110u;
8) QRS width in ms; 9) presence or absence of bundle branch
block defined as a QRS width .100ms, and stratification in
right bundle branch block or left bundle branch block; 10) the
presence or absence of RVS pattern defined as ST-segment
deviation and T-wave inversion in leads V1–V3 [25]; 11) the
presence or absence of left ventricular strain pattern defined as
ST-segment deviation and T-wave inversions in leads V5 and
V6; 12) QT-length in ms and corrected QT-length calculated by
the Bazett formula [26]. RVH was defined by a ratio of R and S
in lead V1 .1.
EUROPEAN RESPIRATORY JOURNAL
VOLUME 37 NUMBER 5
1097
c
PULMONARY VASCULAR DISEASE
D. BONDERMAN ET AL.
Assessments
A detailed medical history including medical conditions known
to be associated with PH was obtained [2, 5, 6, 27]. Associated conditions were collagen vascular disease, HIV infection, history of appetite suppressant intake, splenectomy or
ventriculo-atrial shunt. In addition, exercise capacity measured
by the distance in meters walked in 6 min, and the modified
New York Heart Association (NYHA) class [28] at presentation
were determined in each patient. Blood gas analysis was
performed using a ABL 510 blood gas analyser (Radiometer
Medical ApS, Brønshøj, Denmark).
Serum NT-proBNP was measured utilising the Elecsys proBNP
kit (Roche, Basel, Switzerland). Haemodynamic assessment by
RHC included measurement of cardiac output utilising both the
Fick equation and the thermodilution method. Pre-capillary PH
was diagnosed if mean pulmonary arterial pressure (P̄pa)
exceeded 25 mmHg at rest, and pulmonary capillary wedge
pressure was ,15 mmHg [2].
Statistical analysis
Statistical computations were performed with SPSS (version
15.0; SPSS Inc., Chicago, IL, USA) and SAS (version 9.1; SAS
Institute Inc., Cary, NC, USA). Continuous variables were
described with mean and SD. Groups were compared using the
unpaired t-test. Right-skewed variables were logarithmically
transformed before testing. Nominal variables were described
with counts and percentages, groups were compared with Chisquared or Fisher’s exact test. Wilson’s method was used
to compute confidence intervals (CI) for single proportions.
McNemar’s test was used to compare sensitivities and
specificities of diagnostic decision rules. All reported p-values
are the result of two-tailed tests, and p-values ,0.05 were
considered statistically significant.
Univariable logistic regression models were used to assess
whether TTE, clinical and ECG variables allowed discrimination between pre-capillary PH and no pre-capillary PH. Stepwise selection (forward search) within the three groups (TTE,
clinical and ECG variables) yielded partially independent
variable sets. The clinical and the ECG variable sets were used
to construct a diagnostic decision tree (Classification and
Regression Tree; CART) for pre-capillary PH versus no precapillary PH employing the CHAID (Chi-Squared Automatic
Interaction Detection) method of SPSS and defining diagnostic
branch points and terminal nodes. At each branch point a case
will either branch to the left or to the right based on a test
against a threshold predictor value, and will continue branching in subsequent nodes until a terminal node is reached.
Because of the prognostic implications of a delayed PAH
diagnosis and treatment [1], a false negative diagnosis was
assumed to have far more serious consequences than a false
positive one. Therefore, the class assignment rule was chosen
in a way that the percentage of false negative predictions did
not exceed 4% (one out of 25) of the true positive cases.
To overcome the problem of overoptimistic results, both an
internal and a temporal validation step were added [29].
Internal validation was based on the bootstrap approach [30],
reporting an average of 30 bootstrap samples. Temporal validation was based on the prospective sample.
1098
VOLUME 37 NUMBER 5
RESULTS
Characteristics of the retrospective patient cohort
Patient characteristics are summarised in table 1. According to
RHC results, 187 (74.5%) patients were classified as pre-capillary
PH and 64 (25.5%) as no pre-capillary PH.
In the pre-capillary PH group, 49 patients were eventually
diagnosed with idiopathic or familial PH, or PH associated with
corrected or small (,2 cm) uncorrected atrial septal defects, two
patients had a history of anorexigen intake, six were HIVpositive, 10 female patients suffered from collagen vascular
disease, two had PH in association with Osler’s disease, 13
patients had underlying lung disease [31], 97 patients had
CTEPH, 7 had porto-pulmonary hypertension, and one patient
suffered from pulmonary veno-occlusive disease.
In the no pre-capillary PH group, 53 patients were diagnosed
with post-capillary PH. Of those, 42 patients suffered from PH
due to left ventricular diastolic dysfunction (34 with systemic
arterial hypertension and 10 with significant coronary artery
disease). Nine patients with post-capillary hypertension suffered from PH due to lung disease and/or hypoxaemia [32]. In
11 patients PH was excluded because P̄pa was ,25 mmHg. Of
those, one patient was diagnosed with pulmonary lymphangioleiomyomatosis, one patient had an isolated pulmonary AV
malformation, two patients had manifest hyperthyreoidism,
two patients were diagnosed with unilateral pulmonary artery
occlusion, two patients had atrial septal defect and three
patients had severe isolated tricuspid valve regurgitation.
There were no between-group differences with respect to demographic characteristics, including age and sex. Furthermore, no
differences were encountered in the 6-min walking distances or
serum creatinine levels. However, statistically significant differences between pre-capillary PH and no pre-capillary PH
were found in NYHA class, serum NT-proBNP, Pa,CO2, associated medical conditions, TTE parameters (e.g. Ppa,sys, right
ventricular function and diameter, left ventricular wall thickness and diastolic function), haemodynamic parameters, heart
rate, and ECG characteristics, e.g. P-wave amplitude and axis,
QRS axis, RVH, bundle branch block, RVS pattern and the
corrected QT interval (table 1).
Characteristics of the prospective patient cohort
Of 64 (52.9%) patients diagnosed with pre-capillary PH (table 2),
26 had CTEPH, 18 idiopathic or familial PH or PH associated
with corrected or small uncorrected atrial septal defect, 13
suffered from PH associated with chronic lung disease [31],
three had porto-pulmonary hypertension, two suffered from PH
in association with collagen vascular disease and two had PH
associated with HIV infection.
Of the 57 (47.1%) patients diagnosed with no pre-capillary PH,
32 suffered from post-capillary PH associated with either
chronic lung disease [32] or diastolic left ventricular dysfunction, and 25 had normal pulmonary pressures [16]. Patients with
normal pulmonary pressures suffered from parenchymal or
bronchial pulmonary diseases (n59), collagen vascular disease
(n53), coronary artery disease (n52), unilateral occlusion of the
pulmonary artery (n51), isolated tricuspid regurgitation (n51),
patent arterial duct (n51) or abnormal pulmonary vein drainage
(n51), and seven patients suffered from isolated systemic
hypertension.
EUROPEAN RESPIRATORY JOURNAL
D. BONDERMAN ET AL.
PULMONARY VASCULAR DISEASE
Retrospective clinical, echocardiographic, haemodynamic and electrocardiographic characteristics of patients with
pre-capillary pulmonary hypertension (PH) and no pre-capillary PH
TABLE 1
Subjects n
Pre-capillary PH
No pre-capillary PH
187
64
p-value
Clinical parameter
56.3¡15.8
60.9¡15.5
Male
Age yrs
71 (38.0)
26 (40.6)
0.77
6MWD m
336¡139
367¡117
0.15
NYHA classes III+IV
Serum NT-proBNP pg?mL-1
0.045
123 (65.8)
29 (45.3)
0.005
1348¡1693
478¡1055
,0.001
Serum creatinine mg?dL-1
1.1¡0.3
1.1¡0.3
0.77
Pa,CO2 mmHg
34.5¡5.3
37.5¡4.3
,0.001
Associated medical condition
52 (27.8)
5 (7.8)
,0.001
Ppa,sys mmHg
82.2¡27.2
50.0¡13.4
,0.001
Right atrial diameter mm
58.0¡9.5
54.6¡8.0
0.012
Right atrial pressure mmHg
10.0¡4.0
6.8¡2.4
,0.001
TR severity: moderate + severe
81 (43.3)
17 (26.6)
0.017
Right ventricular dysfunction
108 (61.4)
2 (3.2)
,0.001
Transthoracic echocardiography parameter
Right ventricular diameter mm
43.8¡8.3
35.0¡5.9
,0.001
Left ventricular hypertrophy
48 (28.9)
31 (50.0)
0.005
Left ventricular diastolic dysfunction
79 (47.6)
43 (69.4)
0.004
Haemodynamic parameter
Cardiac output L?min1#
4.5¡1.3
6.0¡1.8
,0.001
49.3¡14.0
23.5¡7.5
,0.001
751.4¡365.3
158.2¡86.2
,0.001
10.4¡5.6
12.6¡5.2
0.007
Heart rate beats?min-1
81.6¡15.3
73.2¡14.9
,0.001
P-wave amplitude mV"
0.21¡0.08
0.15¡0.05
,0.001
64.3¡18.0
56.1¡22.9
0.019
67.3¡64.5
40.5¡43.3
,0.001
P̄pa mmHg
PVR dyn?s?cm-5
Ppcw mmHg
ECG parameter
P-wave axisu
"
QRS axisu
QRS duration ms
102¡20
97¡19
0.14
Right ventricular hypertrophy+
92 (49.2)
7 (10.9)
,0.001
Right bundle branch block
66 (35.3)
4 (6.3)
,0.001
2 (1.1)
8 (12.5)
,0.001
147 (78.6)
5 (7.8)
,0.001
Left ventricular strain
16 (8.6)
7 (10.9)
0.62
QT interval ms
384¡38
390¡45
0.32
Corrected QT interval
434¡34
424¡31
0.041
Left bundle branch block
Right ventricular strain
Data are presented as mean¡ SD or n (%), unless otherwise stated. 6MWD: 6-min walk distance; NYHA: New York Heart Association; NT-proBNP: N-terminal brain
natriuretic peptide; Pa,CO2: arterial carbon dioxide tension; Ppa,sys: systolic pulmonary artery pressure; TR: tricuspid valve regurgitation; P̄pa: mean pulmonary artery
pressure; PVR: pulmonary vascular resistance; Ppcw; pulmonary capillary wedge pressure. #: derived from measurements based on the thermodilution method; ": 19
patients were excluded from p-wave analysis because of atrial fibrillation; +: diagnosed when the ratio of R and S in lead V1 was .1.
p50.007), Pa,CO2 (OR 0.86, 95% CI 0.79–0.93; p,0.001) and
associated medical conditions (OR 3.37, 95%CI 1.04–10.90;
p50.043) remained independent discriminative factors. ECG
variables with the strongest diagnostic accuracy were heart
rate (OR 1.05, 95% CI 1.02–1.08; p,0.001) and RVS (OR 52.93,
95% CI 17.27–162.18; p,0.001).
Predictors of diagnosis
Based on retrospective patient data, univariable (table 3) and
three separate multivariable logistic regression models
(table 4) were constructed for TTE, clinical parameters and
ECG variables. Ppa,sys (OR 1.06, 95% CI 1.03–1.09; p,0.001),
right ventricular dysfunction (OR 10.28, 95% CI 2.18–48.44;
p50.003) and the absence of left ventricular hypertrophy
(OR 0.34, 95% CI 0.15–0.75; p50.008) were identified as
independent predictors of pre-capillary PH. Of the clinical
variables tested, serum NT-proBNP (OR 2.01, 95% CI;
Accuracy of CART
Because all patients with Ppa,sys o36 mmHg and a suspicion
of PH were referred for RHC, the sensitivity of TTE was 100%
EUROPEAN RESPIRATORY JOURNAL
VOLUME 37 NUMBER 5
1099
c
PULMONARY VASCULAR DISEASE
D. BONDERMAN ET AL.
Prospective clinical, echocardiographic, haemodynamic and electrocardiographic characteristics of patients with precapillary pulmonary hypertension (PH) and no pre-capillary PH
TABLE 2
Subjects n
Pre-capillary PH
No pre-capillary PH
64
57
p-value
Clinical parameter
59.8¡15.6
63.7¡11.9
Male
Age yrs
27 (42.2)
23 (40.4)
0.84
6MWD m
325¡126
330¡126
0.83
NYHA classes III+IV
Serum NT-proBNP pg?mL-1
0.13
49 (76.6)
29 (45.3)
0.004
3648¡6541
1489¡3518
,0.001
Serum creatinine mg?dL-1
1.2¡0.6
1.1¡0.2
0.23
Pa,CO2 mmHg
36.1¡7.2
38.4¡5.3
0.049
Associated medical condition
20 (31.3)
3 (5.3)
,0.001
,0.001
Transthoracic echocardiography parameter
Ppa,sys mmHg
82.6¡24.3
55.2¡16.3
Right atrial diameter mm
58.7¡10.9
59.1¡11.5
0.870
Right atrial pressure mmHg
9.5¡3.2
8.4¡3.5
0.074
TR severity: moderate + severe
36 (56.3)
22 (38.6)
0.016
Right ventricular dysfunction
42 (65.6)
2 (3.5)
,0.001
,0.001
Right ventricular diameter mm
44.0¡9.2
38.2¡6.9
Left ventricular hypertrophy
24 (37.5)
30 (52.6)
0.39
Left ventricular diastolic dysfunction
31 (48.4)
42 (73.7)
0.001
Haemodynamic parameter
Cardiac output L?min-1#
4.2¡1.1
5.0¡1.4
0.001
46.8¡13.4
28.4¡11.0
,0.001
736.8¡332.0
107.2¡121.6
,0.001
10.9¡5.1
16.5¡8.4
,0.001
Heart rate beats?min-1
85.7¡14.6
77.3¡18.6
0.007
P-wave amplitude mV"
0.19¡0.07
0.15¡0.06
0.004
61.0¡29.7
65.3¡15.2
0.43
75.3¡66.7
48.3¡49.5
0.02
P̄pa mmHg
PVR dyn?s?cm-5
Ppcw mmHg
ECG parameter
P-wave axisu
"
QRS axisu
QRS duration ms
103¡20
99¡21
0.36
Right ventricular hypertrophy+
37 (57.8)
9 (15.8)
,0.001
Right bundle branch block
12 (18.8)
8 (14.0)
0.49
0 (0.0)
1 (1.8)
0.47
Left bundle branch block
Right ventricular strain
41 (64.1)
10 (17.5)
,0.001
Left ventricular strain
12 (18.8)
8 (14.0)
0.49
QT interval ms
380¡39
384¡41
0.29
Corrected QT interval
433¡31
426¡32
0.053
Data are presented as mean¡standard deviation or n (%), unless otherwise stated. 6MWD: 6-min walk distance; NYHA: New York Heart Association; NT-proBNP:
N-terminal brain natriuretic peptide; Pa,CO2: arterial carbon dioxide tension; Ppa,sys: systolic pulmonary artery pressure; TR: tricuspid valve regurgitation; P̄pa: mean
pulmonary artery pressure; PVR: pulmonary vascular resistance; Ppcw; pulmonary capillary wedge pressure. #: derived from measurements based on the thermodilution
method; ": 27 patients were excluded from p-wave analysis because of atrial fibrillation; +: diagnosed when the ratio of R and S in lead V1 was .1.
and the specificity was 0.0%. For a more specific noninvasive diagnostic decision tree, all independent clinical and
ECG parameters that had been identified in the multivariable logistic regression models (table 4) were fed into
the CART algorithm. The model automatically selected RVS
and NT-proBNP for primary decision tree construction
(fig. 1).
CART had a sensitivity of 99.4% and a specificity of 40.7%.
Because these figures may be too optimistic as they are
computed from data the tree had been derived from, two
1100
VOLUME 37 NUMBER 5
validation steps were added. Internal validation by the
bootstrap technique was performed on the original data set
resulting in a sensitivity of 97.9% (95% CI 94.5–99.2%) and a
specificity of 17.3% (95% CI 9.5–29.5%). Temporal validation
was performed on data from 121 prospectively recruited
patients and yielded a sensitivity of 100% (95% CI 94.3–100%)
and a specificity of 19.3% (95%CI: 11.1–31.3%) (table 5). While
sensitivity was unchanged (100% versus 100%; p51.0), specificity had improved from 0.0% according to current clinical
practice to 19.3% (p50.0009).
EUROPEAN RESPIRATORY JOURNAL
D. BONDERMAN ET AL.
TABLE 3
PULMONARY VASCULAR DISEASE
Clinical, echocardiographic and ECG
parameters to predict pre-capillary pulmonary
hypertension in the retrospective cohort
(univariable logistic regression)
OR (95%CI)
Clinical, echocardiographic and ECG
parameters to predict pre-capillary pulmonary
hypertension in the retrospective cohort
(multivariable logistic regression)
TABLE 4
p-value
Clinical parameter
OR (95% CI)
p-value
Clinical parameter
Age yrs
0.98 (0.96–1.00)
0.047
Serum NT-proBNP ng?mL-1#
2.01 (1.21–3.33)
0.007
Male
0.89 (0.50–1.600
0.71
Pa,CO2 mmHg
0.86 (0.79–0.93)
,0.001
6MWD km#
0.17 (0.02–1.93)
0.15
Associated medical condition
3.37 (1.04–10.90)
NYHA classes III+IV
2.32 (1.30–4.13)
0.004
TTE parameter
Serum NT-proBNP ng?mL-1#
2.05 (1.31–3.22)
0.002
Ppa,sys mmHg
Pa,CO2 mmHg
0.89 (0.84–0.95)
,0.001
Associated medical condition
4.55 (1.73–11.96)
0.002
TTE parameter
1.06 (1.03–1.09)
0.043
,0.001
Right ventricular dysfunction
10.28 (2.18–48.44)
0.003
Left ventricular hypertrophy
0.34 (0.15–0.75)
0.008
ECG parameter
Ppa,sys mmHg
1.08 (1.05–1.10)
Right atrial diameter mm
1.04 (1.01–1.08)
Right atrial pressure mmHg
1.32 (1.19–1.47)
TR severity moderate + severe
2.36 (1.26–4.42)
Right ventricular dysfunction
48.44 (11.47–204.61)
,0.001
0.0113
Heart rate beats?min-1
1.05 (1.02–1.08)
Right ventricular strain
52.93 (17.27–162.18)
,0.001
,0.001
,0.001
0.008
NT-proBNP: N-terminal brain natriuretic peptide; Pa,CO2: arterial carbon dioxide
,0.001
tension; TTE: transthoracic echocardiography; Ppa,sys: systolic pulmonary
artery pressure. #: units of measurement were adjusted for better readability
Right ventricular diameter mm
1.18 (1.12–1.24)
,0.001
Left ventricular hypertrophy
0.41 (0.22–0.74)
0.003
Left ventricular diastolic dysfunction
0.40 (0.22–0.75)
0.004
Heart rate beats?min-1
1.04 (1.02–1.06)
,0.001
P-wave amplitude mV#
1.01 (1.01–1.02)
,0.001
P-wave axisu
1.02 (1.01–1.04)
0.009
QRS axisu
1.01 (1.00–1.01)
0.003
QRS duration ms
1.01 (1.00–1.03)
0.14
Right ventricular hypertrophy
7.89 (3.42–18.19)
,0.001
Right bundle branch block
8.18 (2.85–23.51)
,0.001
Left bundle branch block
0.08 (0.02–0.37)
of confidence intervals.
ECG parameter
Right ventricular strain
43.37 (16.32–115.26)
0.001
,0.001
Left ventricular strain
0.76 (0.30–1.95)
0.57
Corrected QT interval
1.01 (1.00–1.02)
0.043
and mortality [2, 34] has supported early recognition and
treatment [20, 35]. Moreover, screening of asymptomatic
individuals at increased risk for PH has been recommended.
While invasive haemodynamic assessment has been considered
the diagnostic gold standard in PH, TTE is the recommended
screening tool [20, 34]. TTE is noninvasive, broadly available
and most studies report a high correlation of 0.57–0.93 between
TTE and invasive measurements of pulmonary arterial systolic
pressures [2]. However, elevated Ppa,sys may result from either
pre-capillary PH or post-capillary PH, which cannot be safely
distinguished by TTE alone. In contrast to post-capillary PH
6MWD: 6-min walk distance; NYHA: New York Heart Association; NT-proBNP:
N-terminal brain natriuretic peptide; Pa,CO2: arterial carbon dioxide tension; TTE:
Patients with clinical suspicion of pre-capillary PH
and Ppa,sys ≥36 mmHg
transthoracic echocardiography; Ppa,sys: systolic pulmonary artery pressure;
TR: tricuspid valve regurgitation.
#
: units of measurement were adjusted for
better readability of confidence intervals.
DISCUSSION
According to current guidelines [2], a significant proportion of
patients referred for unexplained dyspnoea are undergoing invasive pressure measurements. To test the accuracy of
standard noninvasive diagnostic tools, we analysed data sets
from 251 retrospective and 121 prospective patients who
underwent RHC for evaluation of PH. The main finding of our
study was that ECG and NT-proBNP in addition to TTE suffice
to predict significant pre-capillary pulmonary vascular disease
at a level of sensitivity of 100% and specificity of 19.3%. In
practice, based on the CART algorithm, RHC can be safely
withheld in 9% of patients with elevated echo Ppa,sys and
clinical PH suspicion without overlooking true PH cases.
ECG
No RVS
RVS
Pre-capillary PH likely
BNP
≤80 pg·mL-1
Pre-capillary PH excluded
FIGURE 1.
>80 pg·mL-1
Pre-capillary PH likely
PAH, which is one of the major causes of pre-capillary PH, is a
severe condition with serious prognosis [33]. The availability of
potent vasodilator therapies that positively impact on morbidity
pressure; BNP: brain natriuretic peptide; RVS: right ventricular strain.
EUROPEAN RESPIRATORY JOURNAL
VOLUME 37 NUMBER 5
Pulmonary hypertension (PH) diagnosis tree based on the
Classification and Regression Tree algorithm. Ppa,sys: systolic pulmonary artery
1101
c
PULMONARY VASCULAR DISEASE
TABLE 5
D. BONDERMAN ET AL.
Prospective validation of the Classification and Regression Tree algorithm
Pre-capillary PH diagnosed by RHC
RVS
No RVS and NT-proBNP
No RVS and NT-proBNP
.80 pg?mL-1
f80 pg?mL-1
Total
Yes
41
23
0
No
11
35
11
64
57
Total
52
58
11
121
Data are presented as n and results reflect the true predictive power of the decision tree because the prospectively recruited patients did not contribute to decision tree
construction. A total of 121 patients were studied. Right ventricular strain (RVS) pattern was present in 52 patients, 41 of whom were diagnosed with pre-capillary
pulmonary hypertension (PH) by right heart catheter (RHC). Patients without RVS pattern on ECG (n569) were further dichotomised according to serum N-terminal brain
natriuretic peptide (NT-proBNP) levels. 58 patients displayed serum NT-proBNP levels .80 pg?mL-1; of these 23 were diagnosed with pre-capillary PH. In contrast, none
of the 11 patients with serum NT-proBNP f80 pg?mL-1 were diagnosed with pre-capillary PH.
[15], pre-capillary vascular disease is a rare condition [7]. These
circumstances have recently led to numerous invasive procedures for exclusion. To narrow the grey zone that is blurring the
distinction between pre- and post-capillary disease by echocardiography, current recommendations propose an invasive
diagnostic work-up in patients with echocardiographic Ppa,sys
values .50 mmHg [36]. According to our assessments, this
diagnostic approach would have substantially increased specificity to 42.1%; however, at the cost of overlooking 6.2% of true
pre-capillary PH cases.
Although of limited value, standard ECG has diagnostic [37]
and prognostic [38] potential in PH. Despite the fact that postcapillary PH may cause an RVS pattern on ECG [39], RVS
remained the strongest predictor of pre-capillary PH in our
study. RVS depicts right ventricular electric repolarisation, and
appears to be a sensitive and immediate marker of right
ventricular strain. The presence of RVS correctly identified
78.8% of pre-capillary PH cases. In addition, several noninvasive prognostic parameters have been established over the
years, e.g. the 6-min walk distance [40], NYHA functional class
[1], Pa,CO2 [23], associated medical conditions [27, 41–43], and
serum NT-proBNP levels [44]. NT-proBNP is released from
both cardiac ventricles in response to increased wall tension
and is elevated in PH correlating well with pulmonary
vascular resistance [45]. Its discriminative accuracy with
respect to cardiac versus noncardiac causes of dyspnoea has
been described previously [46], and is primarily based upon its
excellent negative predictive value of 96% [47]. The combination of RVS and NT-proBNP, one being a strong positive, the
other a strong negative predictor, confers clinical usefulness.
A main limitation of our study is its single centre design.
A centre-specific bias with respect to diagnostic procedures, in
particular TTE, cannot be excluded. Different referral patterns
may influence proportions of PH versus non-PH cases. However, in contrast to positive and negative predictive values,
sensitivities and specificities for the detection of PH are
independent of the prevalence of healthy individuals in a
given cohort.
Taken together, we would like to emphasise that RHC remains
the gold standard for the evaluation of PH [11–13]. However, it
is widely accepted that a diagnostic procedure that is primarily
undertaken to rule out a disease is ideally noninvasive. In a
1102
VOLUME 37 NUMBER 5
selected patient population referred for dyspnoea and echocardiographic suspicion of PH, integration of the decision tree
subsequent to TTE helps to avoid unnecessary RHCs in 9% of
cases while not missing a single true PH case.
SUPPORT STATEMENT
This study received financial support from by the European
Commission under the 6th Framework Programme (Contract No
LSHM-CT-2005-018725, PULMOTENSION to I.M. Lang). This publication reflects only the author’s views and the European Community is
in no way liable for any use that may be made of the information
contained therein. The research was supported by the Austrian
fellowship grant FWFL 513-B11 to D. Bonderman and the Hans und
Blanca Moser Stiftung to A.M. Martischnig.
STATEMENT OF INTEREST
None declared.
REFERENCES
1 D’Alonzo GE, Barst RJ, Ayres SM, et al. Survival in patients with
primary pulmonary hypertension. Results from a national
prospective registry. Ann Intern Med 1991; 115: 343–349.
2 Galie N, Torbicki A, Barst R, et al. Guidelines on diagnosis and
treatment of pulmonary arterial hypertension. The Task Force on
Diagnosis and Treatment of Pulmonary Arterial Hypertension
of the European Society of Cardiology. Eur Heart J 2004; 25:
2243–2278.
3 Abenhaim L, Moride Y, Brenot F, et al. Appetite-suppressant drugs
and the risk of primary pulmonary hypertension. International
Primary Pulmonary Hypertension Study Group. N Engl J Med
1996; 335: 609–616.
4 Petitpretz P, Brenot F, Azarian R, et al. Pulmonary hypertension in
patients with human immunodeficiency virus infection.
Comparison with primary pulmonary hypertension. Circulation
1994; 89: 2722–2727.
5 Bonderman D, Jakowitsch J, Adlbrecht C, et al. Medical conditions
increasing the risk of chronic thromboembolic pulmonary hypertension. Thromb Haemost 2005; 93: 512–516.
6 Bonderman D, Wilkens H, Wakounig S, et al. Risk factors for
chronic thromboembolic pulmonary hypertension. Eur Respir J
2009; 33: 325–331.
7 Gaine SP, Rubin LJ. Primary pulmonary hypertension. Lancet 1998;
352: 719–725.
8 Bonderman D, Jakowitsch J, Redwan B, et al. Role for staphylococci in misguided thrombus resolution of chronic thromboembolic pulmonary hypertension. Arterioscler Thromb Vasc Biol 2008;
28: 678–684.
EUROPEAN RESPIRATORY JOURNAL
D. BONDERMAN ET AL.
PULMONARY VASCULAR DISEASE
9 Jais X, Ioos V, Jardim C, et al. Splenectomy and chronic thromboembolic pulmonary hypertension. Thorax 2005; 60: 1031–1034.
10 Lang IM. Chronic thromboembolic pulmonary hypertension: not
so rare after all. N Engl J Med 2004; 350: 2236–2238.
11 Badesch DB, Champion HC, Sanchez MA, et al. Diagnosis and
assessment of pulmonary arterial hypertension. J Am Coll Cardiol
2009; 54: Suppl. 1, S55–S66.
12 Galie N, Hoeper MM, Humbert M, et al. Guidelines for the
diagnosis and treatment of pulmonary hypertension. Eur Respir J
2009; 34: 1219–1263.
13 Galie N, Hoeper MM, Humbert M, et al. Guidelines for the
diagnosis and treatment of pulmonary hypertension: The Task
Force for the Diagnosis and Treatment of Pulmonary Hypertension of the European Society of Cardiology (ESC) and the
European Respiratory Society (ERS), endorsed by the International
Society of Heart and Lung Transplantation (ISHLT). Eur Heart J
2009; 30: 2493–2537.
14 Hoeper MM, Barbera JA, Channick RN, et al. Diagnosis, assessment,
and treatment of non-pulmonary arterial hypertension pulmonary
hypertension. J Am Coll Cardiol 2009; 54: Suppl. 1, S85–S96.
15 Paulus WJ, Tschope C, Sanderson JE, et al. How to diagnose
diastolic heart failure: a consensus statement on the diagnosis of
heart failure with normal left ventricular ejection fraction by the
Heart Failure and Echocardiography Associations of the European
Society of Cardiology. Eur Heart J 2007; 28: 2539–2550.
16 Fisher MR, Forfia PR, Chamera E, et al. Accuracy of Doppler
echocardiography in the hemodynamic assessment of pulmonary
hypertension. Am J Respir Crit Care Med 2009; 179: 615–621.
17 Michelakis ED. Soluble guanylate cyclase stimulators as a
potential therapy for PAH: enthusiasm, pragmatism and concern.
Eur Respir J 2009; 33: 717–721.
18 Hoeper MM, Lee SH, Voswinckel R, et al. Complications of right
heart catheterization procedures in patients with pulmonary
hypertension in experienced centers. J Am Coll Cardiol 2006; 48:
2546–2552.
19 Schenk P, Stiebellehner L, Burghuber OC, et al. [Examination of
pulmonary circulation using right heart catheterization. Position
paper of the Cardiopulmonary Task Force of the Austrian Society
for Pneumology and the Austrian Society for Cardiology]. Wien
Klin Wochenschr 2005; 117: 651–662.
20 McGoon M, Gutterman D, Steen V, et al. Screening, early detection,
and diagnosis of pulmonary arterial hypertension: ACCP evidencebased clinical practice guidelines. Chest 2004; 126: Suppl. 1, 14S–34S.
21 Rich S, Dantzker DR, Ayres SM, et al. Primary pulmonary
hypertension. A national prospective study. Ann Intern Med
1987; 107: 216–223.
22 Fijalkowska A, Kurzyna M, Torbicki A, et al. Serum N-terminal
brain natriuretic peptide as a prognostic parameter in patients
with pulmonary hypertension. Chest 2006; 129: 1313–1321.
23 Hoeper MM, Pletz MW, Golpon H, et al. Prognostic value of blood
gas analyses in patients with idiopathic pulmonary arterial
hypertension. Eur Respir J 2007; 29: 944–950.
24 Dickstein K, Cohen-Solal A, Filippatos G, et al. ESC Guidelines for
the diagnosis and treatment of acute and chronic heart failure
2008: the Task Force for the Diagnosis and Treatment of Acute and
Chronic Heart Failure 2008 of the European Society of Cardiology.
Developed in collaboration with the Heart Failure Association of
the ESC (HFA) and endorsed by the European Society of Intensive
Care Medicine (ESICM). Eur Heart J 2008; 29: 2388–2442.
25 Braunwald E. Braunwald’s Heart Disease: A Textbook of Cardiovascular Medicine. 5th Edn. Philadelphia, Elsevier Health, 2004.
26 Burchell HB. The QT interval historically treated. Pediatr Cardiol
1983; 4: 139–148.
27 Bonderman D, Skoro-Sajer N, Jakowitsch J, et al. Predictors of
outcome in chronic thromboembolic pulmonary hypertension.
Circulation 2007; 115: 2153–2158.
28 Hoeper MM, Oudiz RJ, Peacock A, et al. End points and clinical trial
designs in pulmonary arterial hypertension: clinical and regulatory
perspectives. J Am Coll Cardiol 2004; 43: Suppl. 12, 48S–55S.
29 Koenig IR, Malley JD, Weimar C, et al. Practical experiences on the
necessity of external validation. Statist Med 2007; 26: 5499–5511.
30 Efron B, Tibshirani RJ. An Introduction to the Bootstrap. New
York, Chapman and Hall, 1993.
31 Weitzenblum E, Chaouat A, Canuet M, et al. Pulmonary hypertension in chronic obstructive pulmonary disease and interstitial lung
diseases. Semin Respir Crit Care Med 2009; 30: 458–470.
32 Funk GC, Lang I, Schenk P, et al. Left ventricular diastolic
dysfunction in patients with COPD in the presence and absence of
elevated pulmonary arterial pressure. Chest 2008; 133: 1354–1359.
33 Humbert M, Sitbon O, Chaouat A, et al. Pulmonary arterial
hypertension in France: results from a national registry. Am J
Respir Crit Care Med 2006; 173: 1023–1030.
34 Grunig E, Weissmann S, Ehlken N, et al. Stress Doppler
echocardiography in relatives of patients with idiopathic and
familial pulmonary arterial hypertension. results of a multicenter
European analysis of pulmonary artery pressure response to
exercise and hypoxia. Circulation 2009; 119: 1747–1757.
35 Galie N, Rubin L, Hoeper M, et al. Treatment of patients with
mildly symptomatic pulmonary arterial hypertension with bosentan (EARLY study): a double-blind, randomised controlled trial.
Lancet 2008; 371: 2093–2100.
36 Olschewski H, Hoeper MM, Borst MM, et al. [Diagnosis and
therapy of chronic pulmonary hypertension]. Pneumologie 2006; 60:
749–771.
37 Ahearn GS, Tapson VF, Rebeiz A, et al. Electrocardiography to
define clinical status in primary pulmonary hypertension and
pulmonary arterial hypertension secondary to collagen vascular
disease. Chest 2002; 122: 524–527.
38 Bossone E, Paciocco G, Iarussi D, et al. The prognostic role of
the ECG in primary pulmonary hypertension. Chest 2002; 121:
513–518.
39 Enomoto H. Clinical studies on the ventricular strain pattern
electrocardioogram (I) : clinical statistical studies. Jap Circ J 1961;
25: 1277–1286.
40 Miyamoto S, Nagaya N, Satoh T, et al. Clinical correlates and
prognostic significance of six-minute walk test in patients with
primary pulmonary hypertension. Comparison with cardiopulmonary exercise testing. Am J Respir Crit Care Med 2000; 161:
487–492.
41 Kawut SM, Taichman DB, Archer-Chicko CL, et al. Hemodynamics and survival in patients with pulmonary arterial
hypertension related to systemic sclerosis. Chest 2003; 123:
344–350.
42 Kawut SM, Taichman DB, Ahya VN, et al. Hemodynamics and
survival of patients with portopulmonary hypertension. Liver
Transpl 2005; 11: 1107–1111.
43 Rich S, Shillington A, McLaughlin V. Comparison of survival in
patients with pulmonary hypertension associated with fenfluramine to patients with primary pulmonary hypertension. Am J
Cardiol 2003; 92: 1366–1368.
44 Nagaya N, Nishikimi T, Uematsu M, et al. Plasma brain natriuretic
peptide as a prognostic indicator in patients with primary
pulmonary hypertension. Circulation 2000; 102: 865–870.
45 Pruszczyk P. N-terminal pro-brain natriuretic peptide as an
indicator of right ventricular dysfunction. J Card Fail 2005; 11:
Suppl. 5, S65–S69.
46 Tabbibizar R, Maisel A. The impact of B-type natriuretic peptide
levels on the diagnoses and management of congestive heart
failure. Curr Opin Cardiol 2002; 17: 340–345.
47 Maisel AS, Krishnaswamy P, Nowak RM, et al. Rapid measurement of B-type natriuretic peptide in the emergency diagnosis of
heart failure. N Engl J Med 2002; 347: 161–167.
EUROPEAN RESPIRATORY JOURNAL
VOLUME 37 NUMBER 5
1103